Highly Efficient Perovskite Solar Cells with Substantial Reduction of Lead Content

Despite organometal halide perovskite solar cells have recently exhibited a significant leap in efficiency, the Sn-based perovskite solar cells still suffer from low efficiency. Here, a series homogeneous CH3NH3Pb(1−x)SnxI3 (0 ≤ x ≤ 1) perovskite thin films with full coverage were obtained via solvent engineering. In particular, the intermediate complexes of PbI2/(SnI2)∙(DMSO)x were proved to retard the crystallization of CH3NH3SnI3, thus allowing the realization of high quality Sn-introduced perovskite thin films. The external quantum efficiency (EQE) of as-prepared solar cells were demonstrated to extend a broad absorption minimum over 50% in the wavelength range from 350 to 950 nm accompanied by a noteworthy absorption onset up to 1050 nm. The CH3NH3Pb0.75Sn0.25I3 perovskite solar cells with inverted structure were consequently realized with maximum power conversion efficiency (PCE) of 14.12%.

Herein, the CH 3 NH 3 Pb (1−x) Sn x I 3 (0 ≤ x ≤ 1) perovskite thin films were prepared by a two-step solution-processed technique that was able to control the crystallization rate of CH 3 NH 3 SnI 3 , which solved the problem of uncontrollable crystal growth due to its habitual nature of easier crystallization even at room temperature 17 . In our previous study, we employed DMSO to complex PbI 2 in the case of pure CH 3 NH 3 PbI 3 20 , by which, the obtained PbI 2 •(DMSO) x (0 ≤ x ≤ 1.86) complexes were proved to be capable of retarding the crystallization of PbI 2 . Here, the DMSO was used to complex the PbI 2 and SnI 2 separately via solvent engineering. We carefully studied the influence of precursor concentration of PbI 2 •(DMSO) x and SnI 2 •(DMSO) x on the thin films and the corresponding solar cells. The Sn precursor solution varied from 0 to 80 μ L with a total volume (80 μ L) of Sn and Pb as mentioned in the experimental section. The corresponding values of x in CH 3 NH 3 Pb (1−x) Sn x I 3 were in the range from 0 to 1. It should be noted that the insitu complex thin films were amorphous and the corresponding XRD peaks were not able to be detected prior to post-annealing treatment. As shown in Supplementary Fig. 1 and Supplementary Fig. 2a, in the case of x = 0, the PbI 2 •(DMSO) x complex was formed as expected, which was consistent with our previous report 20 . While the Sn was introduced, the peaks that assigned to be SnI 2 •(DMSO) x complex were also observed ( Supplementary Fig. 1). To further clarify the presence of the intermediates, we carried out the FTIR characterizations of the as-prepared intermediate layers ( Supplementary Fig. 2), which were deposited from DMSO/dimethylformamide (DMF) solution before and after annealing. The FTIR spectra show the characteristic C− S and C− O stretching vibrations from the Sn 2+ -and/or Pb 2+ -coordinated DMSO solvent at 960 and 1012 cm −1 , and C = O stretching vibrations from the Sn 2+ -and/or Pb 2+ -coordinated DMF solvent at 1389 and 1688 cm −1 . After heated at 60 °C for 10 min, the stretching vibrations of C = O were disappeared while the C− S and C− O stretching vibrations were still exist. Combined with the XRD results ( Supplementary Fig. 1), we can come to a conclusion that SnI 2 (DMSO) x and/or PbI 2 (DMSO) x were formed at relatively low temperatures, respectively 18 . Upon annealing at 140 °C, we found that the characteristic modes of the DMSO molecule disappeared, thus confirming the complete removal of the intermediate compound. Additionally, we found that the grain size of Sn-based complex had a significant enhancement in comparison to the one of PbI 2 -based complex, which was associated with the properties of easy crystallization and fast growth rate for SnI 2 ( Supplementary  Fig. 3a,b). We assumed that the SnI 2 dominated the complexed process in the copresence of SnI 2 and PbI 2 that dissolved in DMSO and DMF precursor solution. There was no any complex detected once the solvent without DMSO in the case x = 1 (Supplementary Fig. 1 and Fig. 3d). The highly crystallized SnI 2 single crystal film with lower coverage prohibited the further process to fabricate solar cells.
As expected, the copresence of PbI 2 •(DMSO) x and SnI 2 •(DMSO) x complexes were demonstrated to slow down the crystallization of CH 3 NH 3 Pb (1−x) Sn x I 3 (0 ≤ x ≤ 1) perovskite thin films. Figure 1 displays the XRD evolution of prepared CH 3 NH 3 Pb (1−x) Sn x I 3 (0 ≤ x ≤ 1) thin films while the involved volume of Sn precursor solution varied from x = 0 to x = 1. Prior to XRD characterizations, we simulated the XRD patterns of x = 0 and x = 1, which represented the CH 3 NH 3 PbI 3 and CH 3 NH 3 SnI 3 , respectively. The simulated results revealed that the perovskite phase altered from tetragonal phase (x = 0, CH 3 NH 3 PbI 3 , P4 mm (α -phase)) to cubic phase (x = 1, CH 3 NH 3 SnI 3 , I4 cm (β -phase)). As shown in Fig. 1, the (110) peak of perovskite gradually moved from 14.1° to 14.2° corresponding to (100) plane in cubic phase while the Sn introduction amount improved from x = 0 to x = 1, whereas the FTO peaks remained stable. Additionally, the two peaks located at 23.5° and 24.5° were respectively assigned to be (211) and (202) in the case of CH 3 NH 3 PbI 3 . With the x value increasing, these two peaks tent to merge to be a single peak corresponding to the (113) plane in the P4 mm space group (CH 3 NH 3 SnI 3 ). Such evolution suggested a distortion of the relative positions of the octahedra along the 4-fold crystallographic axis. It turns out that the involved Sn induced the alteration of the perovskite phase, which is consistent with the previous report 11 .
Meanwhile, the superfluous PbI 2 /SnI 2 appeared when the Sn involved inside the perovskite thin films as shown in Fig. 1a. It can also be certified by the morphology of perovskite thin film as shown in SEM images  Fig. 4). As previously reported [21][22][23] , the sporadic formation of PbI 2 /SnI 2 species in the grain boundaries might give rise to a successful passivation that forms energy barriers to prevent excitons from the surface defects and/or traps states.
We then proposed the potential mechanism in the kinetic process of perovskite crystal growth in the presence of DMSO as shown in Fig. 2. Here, the PbI 2 /SnI 2 can be intercalated by DMSO solvent mediation, which gives rise to the formation of PbI 2 /(SnI 2 )•(DMSO) x complexes (Fig. 2a). During the complexing process, the DMSO solvent is advantageous to the formation of amorphous mirror-like film as shown in Supplementary Fig. 3. Meanwhile, there might be a competed relationship while the PbI 2 /SnI 2 complexes with DMSO. The SnI 2 is supposed to govern the complexing process owing to its higher activation energy. Afterward, the presence of CH 3 NH 3 I (MAI) will exchange DMSO due to its higher affinity ability toward PbI 2 /SnI 2 in comparison to DMSO, which allows the formation of CH 3 NH 3 Pb (1−x) Sn x I 3 (0 ≤ x ≤ 1) thin films with highly dense and full coverage. As discussed above, the phase of perovskite thin film strongly depends on the x value. When x < 0.5, the structure is assigned to be tetragonal phase, whereas it is cubic phase once x ≥ 0.5 (Fig. 2b). Figure 3a exhibits the UV-vis absorbance of series CH 3 NH 3 Pb (1−x) Sn x I 3 (0 ≤ x ≤ 1) thin films. With the Sn content increased, the optical absorbance band-edge of thin film clearly moved toward infrared region in consideration of the narrowed bandgap of CH 3 NH 3 SnI 3 . Figure 3b displays the PL spectra of series CH 3 NH 3 Pb (1−x) Sn x I 3 (0 ≤ x ≤ 1) thin films. In the case of x = 0, the PL spectrum was located at 775 nm, which was assigned to the recombination of electron-hole pair for the typical CH 3 NH 3 PbI 3 thin film 24,25 . Again, the intermediate perovskite thin films exhibited an infrared shift with Sn content increasing. The corresponding PL spectrum was consequently fixed at 985 nm. In the case of x = 1, i.e. the case of pure CH 3 NH 3 SnI 3 thin film, unexpectedly, the PL spectrum shift to shorter wavelength (955 nm) in comparison to the case of x = 0.75. Similar phenomenon was also observed in the previous report 11 , which was associated with the tunable bandgap for the intermediate-alloyed It was demonstrated that the Sn 2+ was easily oxidized into Sn 4+ , which likely cause the instability of Sn-based perovskite solar cells 19 . To study the stability of as-prepared CH 3 NH 3 Pb (1−x) Sn x I 3 (0 ≤ x ≤ 1) thin films, we carried out its XPS characterizations (Fig. 3c). As expected, we did not detect any signal from the Sn element in the sample x = 0. Once the Sn involved, the peak that assigned to Sn element appeared, which was fitted into two main peaks, the corresponding peaks of 485.8 eV and 486.8 eV were associated with the Sn 2+ and Sn 4+ , respectively. There was a tendency that the Sn 2+ decreased, whereas the Sn 4+ increased while enhancing the Sn content in the perovskite thin films. It appeared that the content of Pb is able to stabilize Sn in its 2+ state somehow. Likewise, the O peaks located at 530.5 eV that were fitted and assigned to SnO 2 , which tent to appear and strengthen with the Sn introduction amount increasing ( Supplementary Fig. 5). This variation further certified that the CH 3 NH 3 Pb (1−x) Sn x I 3 (0 ≤ x ≤ 1) thin films preferred to be oxidized while the Pb 2+ content decreased as we mentioned previously. As expected, the peak intensity of Sn element gradually enhanced, whereas the one of Pb displayed a decreased trend with the Sn introduction amount increasing. Besides, the peak positions of Pb and I elements had a small shift while varying the x value, such fluctuation implies that the coordination environment of Pb and/or I was altered when Sn 2+ was introduced inside the crystal lattice.  Figure 4 shows the representative SEM images and the corresponding EDS element mapping of Pb, Sn and I in the CH 3 NH 3 Pb 0.75 Sn 0.25 I 3 film. Apparently, the Sn, Pb and I were homogeneously distributed throughout the film without any obvious phase separation. We also quantified the atomic ratio of Sn over Sn+ Pb as the function of Sn introduction amount by EDS analysis as shown in Fig. 4c. We note that significantly less Sn was presented in the final perovskite than added to the precursor in the intermediary concentrations. The loss of the Sn in the final perovskite films was likely associated with the lower solubility of Sn 2+ in isopropanol. Particularly, in such two-step process, while the CH 3 NH 3 I precursor solution in isopropanol was dropped on the SnI 2 /PbI 2 film, a little amount of SnI 2 was dissolved and thrown off with the spin coating process. Supplementary Fig. 4 clearly demonstrates the morphology evolution of CH 3 NH 3 Pb (1−x) Sn x I 3 thin films from x = 0 to x = 1. With the Sn content increasing, the grain sizes of perovskite tent to be enlarged. However, the roughness and coverage of the film suddenly become very poor in the case of x = 1, which further certifies that the presence of lead is able to stabilize the Sn perovskite thin films as mentioned above. Likewise, considering that the growth rate of SnI 2 -based complex was faster than that of PbI 2 -based complex, the cubic phase of CH 3 NH 3 Pb (1−x) Sn x I 3 (x ≥ 0.5) thin films were easier to obtain. In such two-step process, we can immediately obtain the cubic Sn-rich perovskite film once spin coating MAI precursor solution at room temperature, which usually gave rise to the film with bigger crystal size but a lower coverage ( Supplementary Fig. 4). In this regard, the current two-step procedure was favorable to prepare the Pb-rich and Sn-poor perovskite films. One-step solution-process with the same precursor solution might be more suitable to prepare Pb-free perovskite films.

Photovoltaic performances of CH 3 NH 3 Pb (1−x) Sn x I 3 (0 ≤ x ≤ 1) perovskite solar cells.
Perovskite solar cells with an inverted structure of FTO/PEDOT:PSS/Perovskite/PCBM/BCP/Ag were then fabricated by the series CH 3 NH 3 Pb (1−x) Sn x I 3 (0 ≤ x ≤ 1) thin films (Supplementary Fig. 6). The use of PEDOT:PSS as hole transport material (HTM) can effectively avoid deteriorating the cells by the commonly used HTM including spiro-OMeTAD, lithiumbis(trifluoromethylsulfonyl) imide salt, and 4-tert-butylpyridine. Figure 5a gives the I-V curves of corresponding cells. The CH 3 NH 3 PbI 3 solar cell shows a decent fill factor (FF) of 80.9%, short-circuit photocurrent density (J sc ) of 20.98 mAcm −2 , an open-circuit voltage (V oc ) of 0.918 V, yielding a PCE of 15.58% under AM1.5G solar illumination. While the Sn species were introduced inside the perovskite thin film, the V oc of corresponding cell decreased, whereas the J sc had a notable enhancement in virtue of the extending absorbed edge. The decreased V oc was ascribed to the lower conduction band edge with decreasing Pb content 26 . After carefully optimized the technic parameters, the maximum PCE of CH 3 NH 3 Pb 0.75 Sn 0.25 I 3 perovskite solar cells can rise up to 14.12%, which, to our knowledge, is the highest reported value so far. In the present study, x = 0.25 appears to be the optimum condition by means of banlancing the tradeoff between Pb-content and photovoltaic performances. We then characterized the I-V curves of CH 3 NH 3 PbI 3 and CH 3 NH 3 Pb 0.75 Sn 0.25 I 3 by using 55 cells and gave the statistics on photovoltaic results, forward and backward scan as shown in Table 1. The results show   that both CH 3 NH 3 PbI 3 and CH 3 NH 3 Pb 0.75 Sn 0.25 I 3 cells have relatively lower hysteresis effects. Likewise, we futher checked the stability under illumination of the CH 3 NH 3 PbI 3 and CH 3 NH 3 Pb 0.75 Sn 0.25 I 3 cells, which was shown in Supplementary Fig. 7. In comparision to pure Pb-based device, more or less Sn-involved inside the perovskite thin films would lead to inferior stability of the corresponding cells, which was consistent with the previous report 18 . The oxidation of Sn 2+ in air might be the main cause that quenched the photovoltaic poerfomances of the Sn-based solar cells. Careful encapsulation technology is under study and would resolve the instability of Sn-based solar cell. It is worthy of noting that the PCE of cells with different amount of Sn species here overall outperformed the ones with same Sn content in previous report 11 . As shown in Fig. 5b, the external quantum efficiency (EQE) of series CH 3 NH 3 Pb (1−x) Sn x I 3 (0 ≤ x ≤ 1) solar cells were demonstrated to cover the whole visible spectrum and realize a broad absorption minimum over 50% from 350 to 950 nm accompanied by a noteworthy absorption onset up to 1050 nm. It should be mentioned that the CH 3 NH 3 SnI 3 solar cell had an ultraviolet-shift in comparison to the case of CH 3 NH 3 Pb 0.25 Sn 0.75 I 3 , which was consistent with the PL characterizations. We calculated the band gap of CH 3 NH 3 Pb (1−x) Sn x I 3 (0 ≤ x ≤ 1) thin films based on the EQE (Fig. 5c). It clearly pointed out that we can easily tune the bandgap of perovskite thin film between 1.18 and 1.56 eV. The intermediate compounds with x = 0.75 exhibited the smallest bandgap of 1.18 eV. Meanwhile, we integrated the J sc that estimated from EQE (Fig. 5d), which were in consistent with the measured J sc as shown in Fig. 5a.

Discussion
The pure Sn-based perovskite thin film still suffered from the poor perovskite film quality and low coverage, which gave rise to a poor photovoltaic performance. In the present system, we assumed that the pure SnI 2 was not able to effectively complex the DMSO to form the intermediate compound in view of its higher activation energy toward crystallization, thus enabling the SnI 2 to crystallize quickly just as the case without any solvent mediation. Moreover, the presence of Sn 4+ in CH 3 NH 3 SnI 3 thin film might also contribute to the quenched J sc . More studies on the stability of CH 3 NH 3 SnI 3 solar cells, e.g. solvent engineering, professional encapsulation are underway.
In summary, we employed DMSO to intercalate inside the lattice structure of PbI 2 /SnI 2 , which allowed the formation of intermediate complexes of PbI 2 /(SnI 2 )•(DMSO) x , by which, the CH 3 NH 3 Pb (1−x) Sn x I 3 (0 ≤ x ≤ 1) thin films with full coverage and decent grain size can be obtained. The tunable bandgap and high quality of Sn-introduced perovskite thin films facilitated the realization of solar cells with maximum PCE of 14.12%. We believe that our study can shed lights on the realization of highly efficient Pb-free perovskite solar cells.

Methods
Film preparation and device fabrication. The fluorine doped tin oxide (FTO)-coated glass (8 Ω/cm 2 , Nippon) was cleaned with deionized water, acetone and alcohol in an ultrasonic washing unit by turn. After dried under N 2 atmosphere, the glass was treated under oxygen plasma for 10 min. PEDOT:PSS (Heraeus-Clevios P 4083, Xi'an p-OLED) was spun on the as-treated substrate at a speed of 3,500 rounds per minute (r.p.m.). The film was then annealed at 140 °C for 10 min. 170 mg of PbI 2 (Alfa Aesar, 99.9985%) and 138 mg SnI 2 (Aldrich, 99.99%) were dissolved in the mixed solvent containing 200 μ L of DMF (Aldrich, 99.9%) and 40 μ L of DMSO (Aldrich, 99.9%), respectively. After heated at 60 °C for 2 h, a certain amount of prepared PbI 2 and SnI 2 solutions were mixed to form the precursor solution. It's worthy of noting that the PbI 2 solution needed to be filtered prior to mixing. The mixed precursor solution was spun on the PEDOT:PSS layer at 5000 (r.p.m.) for 30 s. After natural evaporation for 10 min to form the PbI 2 (SnI 2 )·(DMSO) x complexes, the MAI solution with the concentration of 60 mg/ml was spun on the substrate at 4000 (r.p.m.) for 30 s. Afterward, the obtained thin films were annealed at 120 °C for 20 min. Exceptionally, the pure PbI 2 -based perovskite film was annealed at 140 °C for 20 min. The [6,6]-phenyl-C61-butyric acid methyl ester (PCBM) dissolved into dichlorobenzene with the concentration of 20 mg/ml was spun on the top of the as-prepared perovskite layers at 2000 (r.p.m.) for 30 s. Finally, 6 nm 2,9-dimethyl-4,7-diphenyl-1,10-Phenanthroline (BCP) and 120nm Ag electrode were sequentially deposited by thermal evaporation. Note that all of the film fabrications were processed in nitrogen-filled glovebox.
Characterizations. The crystal structure was characterized by Bruker D8 Advance X-ray diffractometer (XRD) with CuKα radiation at 40 kV and 40 mA. Field-emission scanning electron microscopy (SEM) was used to characterize the morphology of the obtained thin film. Both top-down and cross-sectional views were obtained using a FEI NovaNanoSEM450. A double beam spectrophotometer (U-4100, Hitachi) equipped with an integrated sphere was used for the UV-vis transmission measurements in the range from 700 to 1100 nm. A Fourier transform infrared spectroscopy was used to collect the FT-IR spectral data in the 4000 cm −1 -400 cm −1 range. The KBr pellet was used for the powdered samples of layer materials scraped from the substrate. X-ray photoelectron spectroscopy (XPS) was measured with a PHI 5300 ESCA Perkin-Elmer spectrometer. All spectra were shifted to account for sample charging using inorganic carbon at 284.80 eV as a reference. The photoluminescence (PL) was carried out under the excitation of a 532-nm continuous-wave laser. The PL signal is sent to a 0.5-m spectrometer through a 50× objective lens and then detected by a liquid nitrogen-cooled CCD detector array. All PL spectrums were normalized. Current-voltage (J-V) characteristics of perovskite solar cells were measured using a semiconductor device analyzer (Keithley 2601B) and a SAN-EI solar simulator (XES-100S1) with an AM 1.5 G spectrum. The illumination power on the sample was adjusted to 1000 W m −2 using a certified reference solar cell (RS-ID-4). A black mask with an aperture (9 mm 2 ) was placed on the top of the device to control the effective electrode area. Both forward and backward scans were performed and the scan speed was fixed at 0.15 V/s. The external quantum efficiency (EQE) of perovskite solar cell device was measured by using spectrum corresponding system (QTesT 1000ADX), with the monochromatic light wavelength ranging from 350 nm to 1100 nm. The monochromatic light beam is produced by a dual grating monochromatic in front of a halogen lamp. A Si reference solar cell with known EQE is used to determine the spectral response of perovskite solar cells.